Seurat Pre-processing of 5m/o and 24m/o mice

##Pre-processing of 5 m/o and 24 m/o mice - muscle snRNA data

This document includes the steps applied for pre-processing and clustering of the snRNA data obtained from [Petrany, et al. 2020] (https://www.nature.com/articles/s41467-020-20063-w#Sec4). The dataset matrices used to create the Seurat object were downloaded from [Synapses Repository] (https://www.synapse.org/#!Synapse:syn21676145/files/).

Setting Up Seurat object and preprocessing

library(Seurat)
library(dplyr)
library(ggplot2)

For 5 m/o mice TA muscle dataset

Loading the matrices and calculating percent of reads that map to mitochondrial genome

#Load the 5m/o filtered matrices 
months5_filtered <- Read10X_h5(filename = "5mo_filtered_feature_bc_matrix.h5", use.names = TRUE)

#Creating the R seurat object for 5 month old mice
fivemonth <- CreateSeuratObject(counts = months5_filtered, project = "A", min.cells = 3, min.features = 200)

#Checking percentage of reads that map to mitochondrial genome
fivemonth[["percent.mt"]] <- PercentageFeatureSet(fivemonth, pattern = "^MT-")

Visualizing QC metrics for the 5m/o datasts. The filtering cutoff has been set reads with features > 200 and < 3200

VlnPlot(fivemonth, features = c("nFeature_RNA", "nCount_RNA"), ncol = 2)

#Visualizing feature-feature relationship using feature scatterplot
featurescatterplot <- FeatureScatter(fivemonth, feature1 = "nCount_RNA", feature2 = "nFeature_RNA")
plot(featurescatterplot)

#Filtering the reads with features < 200 and >3200
fivemonth <- subset(fivemonth, subset = nFeature_RNA > 200 & nFeature_RNA < 3200)

#Normalizing the Data 
fivemonth <- NormalizeData(fivemonth)
#identifying highly variable features
fivemonth <- FindVariableFeatures(fivemonth, selection.method = "vst", nfeatures = 2000)

#identify top 10 highly variable genes
top10 <- head(VariableFeatures(fivemonth), 10) 

#plot variables features 
plot1variable <- VariableFeaturePlot(fivemonth)
plot2variable <- LabelPoints(plot = plot1variable, points = top10, repel = TRUE)
CombinePlots(plots = list(plot1variable, plot2variable))

#Scaling the data
all.genes5mo <- rownames(fivemonth)
fivemonth <- ScaleData(fivemonth, features = all.genes5mo)

#performing linear pca
fivemonth <- RunPCA(fivemonth, features = VariableFeatures(object = fivemonth))
VizDimLoadings(fivemonth, dims = 1:2, reduction = "pca")

DimPlot(fivemonth, reduction = "pca")

DimHeatmap(fivemonth, dims = 1:15, cells = 500, balanced = TRUE)

Clustering cells and using UMAP for dimensionality reduction

#Clustering the cells
fivemonth <- FindNeighbors(fivemonth, dims = 1:12)
fivemonth <- FindClusters(fivemonth, resolution = 0.5)

## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
## 
## Number of nodes: 8197
## Number of edges: 279293
## 
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.8778
## Number of communities: 13
## Elapsed time: 0 seconds

#Running UMAP
fivemonth <- RunUMAP(fivemonth, dims = 1:12)
fivemonth <- RenameIdents(fivemonth, '0' = "Type IIb Myonuclei", '1' = "Type IIx Myonuclei", '2' = "Type IIb Myonuclei", '3' = "Type IIx Myonuclei", '4' = "FAPs", '5' = "Endothelial Cells", '6' = "Myotendinous Junction", '7' = "Smooth Muscle", '8' = "Satellite Cells", '9' = "Immune Cells", '10' = "Smooth Muscle", '11' = "Neuromuscular Junction", '12' = "Subcutaneous Fat")

DimPlot(fivemonth, reduction = "umap", label = TRUE)

fivemonth_muscle <- subset(fivemonth, idents = c("Type IIx Myonuclei", "Type IIb Myonuclei", "Satellite Cells", "Neuromuscular Junction", "Myotendinous Junction"))
fivemonth$CellType <- Idents(fivemonth)

For 24 m/o mice TA muscle dataset